Britain's 40% duty on online gaming will push more players to the black market

Flutter has said it will cut up to 1,400 jobs, about 15.6% of its roughly 9,000 staff in the UK and Ireland, and the company puts the cuts down to Britain's new tax on online gaming. Remote Gaming Duty (RGD) rose from 21% to 40% on 1 April 2026, and the Budget on 28 October could bring a rise in another gambling tax, Machine Games Duty. I think the 40% rate will push more players to the black market, and that banning or punishing a legal market is the wrong way to deal with gambling harm. A medium tax with real regulation, and enforcement that reaches illegal sites, will do more for players and for the Treasury. The examples I lean on are alcohol and cannabis in the United States.

The mechanism comes from the government's own papers. The Office for Budget Responsibility (OBR) assumes that around 90% of the extra duty is passed on to customers through higher prices or lower payouts, and its estimate of how demand responds includes a move to the illicit market. HM Revenue and Customs (HMRC) lists gambling "through the illegal gambling market" among the things players might do. What nobody has done is measure how many players have gone. The Gambling Commission says the data is too thin for a robust estimate of the illegal market's size.

What changed on 1 April 2026, and what comes in 2027?

At Budget 2025, on 26 November 2025, the government announced that RGD "will be increased from 21% to 40% from 1 April 2026", in the words of HMRC's policy paper on the changes. A new duty rate of 25% for remote betting follows on 1 April 2027. UK horseracing is excluded, and spread betting and pool betting are unchanged. Bingo Duty, which stood at 10%, was abolished from 1 April 2026. HMRC expects the changes to "raise over £1 billion per year".

The Chancellor's case in the Budget speech was harm: "Remote Gaming is associated with the highest levels of harm". I don't dispute the harm. The question is what a higher tax does about it.

The duty isn't the operators' only new cost. A statutory levy, separate from the duty and ranging from 0.1% to 1.1%, took effect on 6 April 2025, and the Treasury's costings for the Spring Statement 2025 put it at about £105m in 2025-26, rising to £115m in 2029-30.

Two dates matter next. The government says it will deliver Budget 2026 to Parliament on 28 October, and a rise in Machine Games Duty (MGD), which has bands of 5%, 20% and 25%, has been reported as possible; the section on Budget 2026 below sets out what's reported. The Gambling Commission's next instalment of research on illegal gambling is due in November.

What did the Treasury expect players to do?

The government never claimed the 40% rate would cost it no custom. HMRC's policy paper says that "some individuals could choose to gamble less, switch to alternative gambling activities or gamble through the illegal gambling market". It gives no estimate of jobs, and says the changes are "not expected to have any significant macroeconomic impacts".

The numbers behind that sit in the OBR's costing. Its November 2025 economic and fiscal outlook assumes operators pass on "around 90 per cent" of the extra duty "by increasing prices or reducing payouts". The policy costings put the figure at 89%, "passed on in higher margins". The OBR then applies elasticities of -1.7 for remote gaming and -1.15 for remote betting, based on the 2019 change to Remote Gaming Duty and on the experience of Denmark and Sweden. For remote gaming, an elasticity of -1.7 means demand falls by more than the effective price rises.

The outlook says those elasticities "also capture potential substitution to the illicit market, and substitution between different forms of gambling". The demand fall cuts the yield "by £0.5 billion by 2029-30". The OBR also expects operators to "restructure their product offering to minimise tax costs", which takes off a further £0.1bn, and it describes the behavioural effects overall as "uncertain but are estimated to reduce the yield by around one-third". The costings name the behavioural response as the main uncertainty.

The policy costings show how much of the yield the Treasury expects to lose. Before behaviour, the changes would raise £1,320m in 2026-27, rising to £1,885m in 2030-31. After behaviour, the figures are £810m and £1,155m. The Treasury priced in a large loss of custom, and it names the response as its main uncertainty.

The government's consultation response, based on 143 replies, described the UK as having "a small illegal market". It quotes industry estimates of between 2% and 9%, and it set aside £26m over three years for enforcement. I come back to that sum below.

What does the tax mean for odds, payouts and bonuses?

The tax lands on operators, and the OBR's assumption is that they hand most of it to customers. For a player at a licensed site, that means a worse deal: higher prices or lower payouts, in the OBR's words, and a product restructured to cut the operator's tax bill.

Industry polling points at promotions as a sensitive spot. A YouGov poll for the Betting and Gaming Council (BGC), the industry body, reported on 22 July 2025, found that 28% of regular punters would seek illegal operators if a tax rise forced offers to be withdrawn. A BGC and YouGov poll of 9 June 2025 found that 65% of regular punters think higher taxes would push customers to unregulated sites. Both come from the industry's own body. The second measures what punters think other customers would do in a hypothetical, which is weaker than asking what they'd do themselves.

Sweden offers a view of why players use unlicensed sites. In a summer 2024 survey reported by the Swedish Gambling Authority (Spelinspektionen) in spring 2025, 19% cited better bonuses and 19% better odds or return to player (RTP) as reasons for unlicensed play. The authority says it can't judge actual RTP differences between licensed and unlicensed sites, so these are perceptions. Each is close to a fifth of respondents, and Britain's Phase 3 research ranks access ahead of odds, which I take up next.

Will players go to the black market?

I expect so. The case against me is the Gambling Commission's third phase of research on illegal online gambling, which it published on 20 October with Yonder. It surveyed more than 4,000 people between 1 and 15 May. It found that using illegal sites is a behaviour of "a small minority of all individuals who gamble". The report puts the strongest drivers down to "deliberate, access-related issues", such as sites that aren't on GamStop and fewer ID checks, and doesn't put "searching for better odds" among them. The words "tax" and "duty" appear nowhere in its summary, findings, methodology or conclusions.

That's a serious finding and I accept it for what it measures. Three points limit what it shows against the 40% rate.

The first is timing. Fieldwork ran in the weeks after the rise. The price effect the OBR assumes arrives through operators' pricing decisions after 1 April, so a survey fielded that soon is early, and it asks about the reasons people use illegal sites, which a shift that hasn't yet happened can't show.

The second is scope. A report that never mentions tax can't say tax doesn't matter. Its findings describe the pull of illegal sites and leave open how much a worse licensed product adds to it.

The third is the government's own model. If demand falls by 1.7% for each 1% the effective price rises, as the OBR assumes, then players are sensitive to price at licensed sites. Players that sensitive have somewhere to go. Phase 3 shows friction pushing a minority towards illegal sites. A 90% pass-through adds a second push, because the licensed product costs more or pays less. Phase 3 describes the first, the OBR and HMRC describe the second, and neither claims to describe both.

Operators say the same, and that adds nothing to the evidence. On Budget day, 26 November 2025, Flutter said in its stock-exchange statement that the duty changes would cost it about $320m in its 2026 financial year and $540m in 2027, and that the tax would "hand a big win to illegal, unlicensed gambling operators". A company that stands to lose from the tax has every reason to say so.

How big is Britain's illegal gambling market?

Nobody knows, and the Gambling Commission says so. In its November 2025 report on dwell time it wrote that "insufficient data is available to refine assumptions so that an estimate can be viewed as robust, reliable and fit for public use". Phase 3 describes use as a small minority of people who gamble. The government's consultation response uses the word "small".

The published estimates come mostly from parties with a stake in the answer. The usual measure is channelisation, the share of play that stays with licensed operators.

• Frontier Economics, working for the BGC, estimated in September 2024 that £2.7bn a year is staked on the black market. That is 2.1% of £128bn in stakes. It found that 8.1% of gamblers, about 1.5 million people, had used a black-market site. The survey had 6,423 respondents, but its question on reasons was answered by only 223.
• Landman, writing in October 2024 for Clean Up Gambling, a campaign group, called the reasons finding "based on a subsample of just 223 responses". Landman describes the black market as "just over 2% of the amount staked" and cites Yield Sec at about 1%.
• PwC, also for the BGC, in late 2025 found only "some correlation (R2 = 0.4-0.6)" between tax and black-market share. Trade press summaries of the report put black-market sites at about 5% of UK online activity.
• H2 Gambling Capital (H2GC), in a report hosted by the BGC in April 2026, put offshore gross gaming yield at £200m in 2019 and £685m in 2025. Its 92% onshore share for 2025 leaves 8% offshore, and it forecasts channelisation falling to 85% by 2031. NEXT.io, a trade news site, asked in a June piece whether the threat had been overstated, and on H2's figures put the UK's online channelisation above 91%.

Add the consultation response's industry estimates of 2% to 9%, and the published figures run from about 1% to 9%, on different bases: stakes, online activity and gross gaming yield. Even the critique written for a campaign group quotes a figure near 2% of stakes. The market is small on every account, and on H2GC's account its offshore yield more than tripled in six years. That is the starting point for the argument: a small market that licensed play can lose more of if it offers worse value.

Where does the "£500m to the black market" figure come from?

It comes from a misreading of the OBR. The figure is repeated in political debate as a forecast that £500m will be lost to the black market. The OBR doesn't say that.

The OBR's November 2025 outlook says the fall in demand cuts the yield "by £0.5 billion by 2029-30", and that its elasticities cover substitution to the illicit market and between forms of gambling. The £0.5bn is the whole demand effect, expressed as tax not collected. Players who bet less, players who move to another kind of licensed gambling and players who move to illegal sites all sit inside it. The OBR gives no separate illicit figure.

The BGC's release of 8 December 2025 wrote of "£500 million lost by 2029-30 as consumers switch away from the regulated sector and towards the black market". Entain's ESG material repeats the claim. So do MPs. Gareth Snell, a Labour MP, repeated the "£500 million… leakage" version in Westminster Hall on 2 December 2025, and in the Commons debate of 13 January 2026 several MPs misstated the OBR as saying £500m of leakage to the black market.

I'd rather the case I'm making didn't borrow from it. If the claim were right, the argument would be easier, because £500m would be a measured loss. It isn't measured, and the OBR's wording is a better foundation. The model assumes players respond to price, and it counts illicit substitution as one of the ways they respond.

Who has lost their job, and is the tax to blame?

Jobs and shops have gone, and several operators put it down to tax. Every row below is the company's own account, so every row comes from an interested party.

Operator	What it announced	What it says is behind it
Flutter	Up to 1,400 redundancies, 15.6% of about 9,000 UK and Ireland staff (28,500 worldwide); cost $80m to $90m; Leeds, Dublin, Porto and Cluj-Napoca	The tax
Evoke	About 270 shops closed (third-quarter trading, 23 October 2026); revenue £406m, down 7%	"higher industry taxation"
Betfred	132 shops and more than 600 jobs (iGB, 3 August 2026)	Tax, National Insurance contributions, wages and the economy
Paddy Power	Up to 100 shops and 400 jobs (September 2026)	Tax plus other causes
bet365	About 340 roles (September 2026)	Competition plus "regulatory and tax-related costs"
Entain	About 400 roles (16 September 2026)	Not attributed to tax

Paddy Power's earlier closure of 57 shops in October 2025 was not tied to tax. Rank, another operator, puts the net cost of the duty changes at £40m, which is £46m more RGD less a £6m saving from abolishing Bingo Duty.

The BGC has said that more than 600 shops and 10,250 jobs have been "lost or announced for redundancy", in statements of 14 and 15 October 2026. That tally mixes causes, and not all of it comes from tax.

So the answer is partly. The cuts are real, and tax, wages, National Insurance contributions and competition all lie behind them. Entain's, about 400 roles, isn't attributed to tax at all. HMRC gave no estimate of jobs, so the counts come from the companies and their trade body, and each has a reason to choose its own explanation.

My argument doesn't need every one of those jobs to be the tax's doing. It rests on the mechanism. The job losses show operators under cost pressure, which is the pressure the OBR expects them to pass on.

Can we trust the industry's job forecasts?

Only with the critique alongside. The BGC commissioned EY, and on 8 December 2025 the BGC published the headline: "nearly 17,000" jobs, more than £6bn of stakes diverted and a 140% rise in the black market. Those were scenarios drawn up before the Budget, and EY never modelled the final 40% and 25% package.

For MGD at 40%, BGC releases of 21 September and 14 October 2026 cite what the BGC calls "Independent EY modelling": between 690 and 1,470 shop closures, 28 to 34 casinos, 11,600 to 15,900 jobs, and a Treasury "up to £124 million" worse off. Those are ranges. The tops of them, 15,900 jobs and 1,470 shops, are the ends of a spread and shouldn't be quoted as forecasts.

The IPPR, which argued for higher gambling taxes and so has its own interest, attacked the EY job figures on 5 November 2025. It wrote that "These estimates are seriously flawed", and said the high-elasticity scenario was "developed in discussion with the BGC". Its verdict: "Discussion with an interested party is not an appropriate basis for credible economic modelling". The target was the BGC's headline of 40,000 jobs. That critique predates the "nearly 17,000" release and the MGD ranges, so it doesn't speak to them directly, but the BGC commissioned both sets.

I weigh the forecasts as the IPPR invites: as scenarios from an industry body, built on assumptions the industry helped set. They tell a reader what the industry fears and how it would like the cost framed. They don't tell anyone what will happen. The tax hawks deserve the same scrutiny. The IPPR wanted RGD at 50% and MGD at 50%, raising "just over £3 billion". The SMF asked for RGD at 50% and "up to £2 billion annually" (July 2025). Both were advocating a number too. The government's 40% sits between the industry and the campaigners, and that doesn't make it right.

Is the extra money coming in?

On receipts, the Treasury looks right so far, and that is real counter-evidence. HMRC's statistics commentary, published on 30 September 2026, puts RGD receipts for April to June at £376m, up 22% on a year earlier. Most of that quarter reflects periods before the rise. RGD is accounted for quarterly, with payment due within a month, and HMRC expects the rise to show in receipts from July 2026 onwards. Trade press then reported July. CDC Gaming, citing provisional HMRC data, put July's RGD receipts at nearly £590m against £283m a year earlier, a rise of more than 108%. The figure sits in HMRC's tables rather than its commentary text. Total betting and gaming receipts for April to July were £1,933m, £309m or 19% higher, and RGD made up half of that total.

A rate that rose by about 90% has met roughly a doubling of receipts in the first month that counts. On those figures the taxed base didn't shrink that month. The Treasury's post-behavioural costing for 2026-27 is £810m.

My answer is that receipts and my argument measure different things. RGD is charged on gross gaming yield, what operators keep after paying out winnings. If operators cut payouts, which is the OBR's own mechanism, yield can hold or rise while players get a worse deal, so a rising tax take can sit beside a worsening product. One month is one month, and the Netherlands, below, shows how far a revenue picture can move from its forecast. The black-market question is about where players go, and a player who moves to an unlicensed site appears in none of HMRC's tables.

What happened in the Netherlands, Sweden, Germany and Ontario?

The Netherlands

The Dutch gambling tax went from 30.5% to 34.2% on 1 January 2025 and to 37.8% in 2026. On a page dated 23 June 2026 the Dutch regulator, the Kansspelautoriteit (KSA), set out with the Finance Ministry whether the rise had met its target. The expected extra revenue was €108m for 2025 and €216m for 2026. The outcome was €2m for 2025 and a forecast of €57m for 2026. The KSA puts the shortfall down to several developments: player-protection measures that cut operators' gross gaming result, and possibly the rate rise itself, for instance through the closure of physical venues.

On the effect on market size and channelisation, the KSA says it can't tell: "Het is niet mogelijk om daarover conclusies te trekken, omdat meerdere veranderingen tegelijk plaatsvonden" (It is not possible to draw conclusions about that, because several changes took place at the same time).

The Dutch deposit limits came before the tax rise. Monthly net-deposit thresholds of €700 and €300 took effect on 1 October 2024, three months before the first rate rise. The KSA's 2025 annual report says the fall in channelisation by money, from 51% at the end of 2024 to 49% at the start of 2025, may possibly be explained by players moving to illegal sites to avoid the new player-protection rules, which it calls a worrying development. The KSA's spring 2026 monitoring report, working from revised data, has channelisation by money slipping from 56% to 53% between the two halves of 2025, and calls that the continuation of a slow, longer-term decline.

So the Dutch case doesn't prove that tax pushed players out. The same monitoring report puts channelisation by players at 91% or 95%, depending on how mixed players are counted, and the annual report says about 94%. The Dutch case shows what happens when a government stacks a rate rise on restrictions: the revenue target is missed by a wide margin and the regulator can't say why. Britain now has a version of that combination, with the friction at licensed sites that Phase 3 identifies as a driver of illegal use, and price on top.

Sweden

Sweden is the best case for a medium tax with strong regulation. The Swedish Gambling Authority puts channelisation at 86% for 2023, 85% for 2024 and 84% for 2025 (in a report of 15 June 2026). Yogonet reports that Sweden's rate rose from 18% to 22% on 1 July 2024. The drift in channelisation is a point a year, with no cliff. The 2024 estimate comes from a report of 1 September 2025 that rests on 5,767 respondents and identified 2,032 websites without a Swedish licence.

The product split is the part that matters for British gaming duty. On the authority's traffic indicator, casino channelisation is 72% and betting 96%. Its survey gives 82% and 92%. The indicators disagree, but both put casino-style play lowest, and casino-style play is the product closest to what RGD taxes. Rates and tax bases differ between countries, so I use Sweden for direction: a regulated market at 22% holds around 85%, and the UK's new rate is close to double that.

Germany and Ontario

Ontario shows what a well-run licensed market can hold. The AGCO and iGaming Ontario, using an Ipsos survey of 2,012 people, reported on 21 May 2026 that 91.1% of players wagered on regulated sites, up from 83.7%. That is a rise of 7.4 percentage points, with a margin of ±2.2.

Germany supports my argument only through industry figures. A study commissioned by Germany's regulator, the GGL, put channelisation at 77.03% for 2024, from a self-reported survey of 2,000 people (reported by iGB on 17 March 2026). The GGL itself calls the result "indicative rather than definitive", and the figure is a share of stakes or losses and isn't specific to slots. The Blockchain Research Lab, with GGL funding, put black-market gross gaming revenue at €547m in 2024. The claim that slot channelisation is below 40% traces to the DOCV, an online casino trade body, speaking at ICE in January 2025, where it gave a range of 20% to 40%. No regulator stands behind it.

What the four countries teach together is modest. The Netherlands is a lesson about revenue forecasts and attribution. Sweden and Ontario show that a regulated market can hold the large majority of play. Germany shows how contested the numbers become.

Can the Gambling Commission stop illegal sites?

It can slow them, which is the case for spending more on enforcement. The Commission's disruption report, updated on 12 March 2026, counts 3,140 cease-and-desist and disruption notices since April 2024, with 447,778 URLs referred and 287,961 removed. Across 160 websites there was "a 32 percent decrease in engagement following disruption", and the Commission writes that "GEO blocking and blocks by the registrar appear to be more effective". An earlier speech by the Commission's chief executive, published on 10 November 2025, reported 480 cease-and-desist letters this financial year, 188,297 URLs reported, 104,192 removed and 504 websites disrupted.

A one-third fall in engagement where disruption was used is a real result. It's also a limit: two-thirds of the traffic remained. And one lever has hardly been pulled. The Commission's impact metrics show three cease-and-desist notices to payment providers, and two instances of payment provision removed or disrupted, in 2025-26.

New powers exist on paper. Section 170 and Schedule 18 of the Crime and Policing Act 2026 give the Crown Court orders for suspending the IP addresses and domains linked to unlicensed gambling. The powers were "not in force at Royal Assent", on 29 April 2026, and as of 26 October legislation.gov.uk lists no commencement regulations for them.

The money is small. Ian Murray, the minister, said in Westminster Hall on 2 December 2025 that "£26 million will be used to tackle the black and illegal market", over three years. The duty rise is meant to raise more than £1bn a year. Over the same three years, £26m is less than 1% of what the duty is expected to bring in.

A government that raises over £1bn a year from a regulated market and sets aside about £9m a year to defend it against illegal competitors has put its money in the wrong place. If the 40% rate pushes players out, the response available to the state is a regulator with a thin payment-blocking record and court powers that haven't been switched on. The Department for Culture, Media and Sport's consultation on banning unlicensed gambling sponsorship and advertising across all sectors, including sport, ran from 15 July to 9 September 2026, and it points the right way. It needs the money and the commencement to match.

What could Budget 2026 change?

The Budget on 28 October could change the picture on land-based gambling, where Machine Games Duty applies.

MGD has bands of 5%, 20% and 25%. City AM reported on 23 October that the SMF, which wants higher gambling taxes, is pressing for MGD to be doubled from 20% to 40%, which it says "could raise up to £458m annually". Reports suggest a smaller rise to 25%. As of 26 October, neither has been announced.

Operators have attached forecasts to the 40% version. Rank's chief executive said on 28 September 2026 that if MGD reaches 40%, a third of its estate and about 2,000 jobs would go. Betfred said on 21 September 2026 that 495 shops and 2,575 jobs would be lost. The BGC's EY ranges, given earlier, add 690 to 1,470 shop closures. These are conditional forecasts from interested parties, and the IPPR's critique applies to how they were built.

The black-market mechanism I've argued for is a different matter on machines in shops. A player at a machine in a high street shop can't switch to an offshore site by walking out of the door, and the forecasts above concern shops and casinos. They test the economics of the estate. They say nothing about the black market. If MGD goes up, the argument against it is about whether the Treasury collects what it expects, since EY says it would be up to £124m worse off, and that is a BGC-commissioned figure.

Two things about the RGD argument stand whatever the Chancellor announces. The mechanism is in the OBR's papers already. The enforcement gap stays until the money and the commencement orders arrive.

What do alcohol and cannabis teach about tax and control?

The United States ran both experiments. I made the longer case in my column on Brazil's betting ban, and the short version follows.

Prohibition cut drinking. Jeffrey Miron and Jeffrey Zwiebel, estimating consumption from mortality, mental-health and crime statistics, found it fell sharply at the start of Prohibition to about 30% of its pre-Prohibition level, rose over the next several years to about 60% to 70%, and returned to about its earlier level during the decade after repeal. Angela Dills and Miron found that constitutional prohibition reduced cirrhosis deaths by about 10% to 20%. The cost was a criminal supply. The National Archives' Prologue magazine says that after the first few years "illegal manufacturing, importation, and sale of alcohol became big business, run by organized crime". The country weighed the gain against that and chose regulation. Prologue says there was no return to national prohibition, that the federal government again collected alcohol taxes, and that "a patchwork of local laws regulated aspects of alcohol policy such as Sunday sales, drinking age".

Cannabis is the recent test. Colorado began taxed adult-use sales in 2014. Its Department of Revenue reports $236,449,405 in marijuana tax and fee revenue for 2025 and over $3.1bn since sales began. A study in JAMA Network Open by D. Mark Anderson and colleagues used national high-school survey data from 1993 to 2019, covering about 1.6 million students. It found no association between a state adopting recreational legalisation and current marijuana use, and a lower rate two or more years later.

The same record shows why the rate matters. California's Department of Cannabis Control wrote in March 2025 that "taxes, fees and other compliance costs can make it expensive to produce licensed cannabis". It put the licensed market's share of consumption at about 40%, up from 25% in 2019, and listed "decreasing costs for licensed businesses" among the ways to shrink the illicit market. The state then cut its retail excise tax from 19% to 15% on 1 October 2025, according to its Legislative Analyst's Office.

By a medium tax I mean one high enough to fund the state and the safeguards, and low enough to keep buyers in the licensed market. Alcohol and cannabis support the same two conclusions about gambling. A ban brings a criminal market. A rate that makes the licensed product expensive hands part of the market back to the illicit one, which is what California's regulator describes. The comparison has one limit that favours enforcement spending: alcohol needed physical production and distribution, and online gambling can be supplied remotely, so the tools that worked on a distillery don't carry over without payment blocking and site disruption.

Where does this leave the 40% rate?

The mechanism is in the government's papers: a 90% pass-through, a demand response with illicit substitution inside it, and HMRC's own list of what players might do. What no source has measured is the share of players who've gone. The first full year of receipts will show whether the Treasury's post-behavioural costing, £810m rising to £1,155m, was close.

Two pieces of evidence work against me, and each limits my claim. Phase 3 finds access and friction drive illegal use, and never mentions tax. July's receipts roughly doubled. Neither answers the OBR's assumption that players respond to price, which is the premise on which the Treasury's own costing stands.

My position is that the licensed market should be taxed at a medium rate and policed properly. I'd rather the Treasury had stopped well short of 40% on remote gaming and put a larger share of its extra revenue into payment blocking, commencement of the new court powers and the Gambling Commission's disruption work. A ban or a punitive rate treats the licensed market as the enemy, and the evidence from alcohol and cannabis is that the enemy is the unlicensed one. If data a year on show no rise in illegal-site use, and receipts hold at the post-behavioural costing, I'll have been wrong about the scale. The first official numbers that bear on it come in the Gambling Commission's November research, which brings the first Illegal Incidence Index and illegal-gambling data from the Gambling Survey for Great Britain.